0
\$\begingroup\$

I am redesigning a battery powered system that communicates with other RS485 systems. The battery powered device provides the supply for the other external systems.
In the old design there was isolation on the communication signals using the ADM2587. But the power was not isolated between them. So the system is not fully isolated. See image:
Diagram

Do this make sense? I am not sure if this partial isolation does something. The system has been in the market for a few years working without problems.

Thank you

EDITED: I add the schema of a part of the battery powered system. The external system connects to the 7 pin connector with a cable. schema

\$\endgroup\$
1
  • 1
    \$\begingroup\$ It may make sense but there is not enough details if it does. Please bear in mind that many people incorrectly think RS-485 is two-wire bus that does not require common ground reference, but it is not a two-wire bus, it requires a common reference between transceiver chips, and without any other route, the bus needs three wires. You might simply be lucky and the bus happens to work due to some detail that you don't reveal in your extremely high level block diagram. \$\endgroup\$ Commented Jul 29 at 13:36

5 Answers 5

Reset to default
3
\$\begingroup\$

There is no isolation as the schematic connects the main ground and isolated transceiver grounds together.

So whoever told you that device has an isolated RS-485 interface is wrong. The chip allows making an isolated interface, but the feature is left unused, so there's just the expensive chip and bunch of other expensive components external to it that could be replaced with simpler and cheaper chip to provide a non-isolated interface.

However, by the same accident, the mistake also allows the circuit to work, because RS-485, or RS-422 as per how you are using the chip, requires a common reference potential between the bus interface chips of the devices, which you now have. Common mistake is also to just wire 2 (or in your case 4) data wires between devices, and it may work on your desk when prototyping it, but in actual target application it just does not work.

Also the design is omitting any termination resistors - it might be fine if they are externally provided on the bus, but a proper RS-485/RS-422 interface requires terminating resistors.

Basically, if you are the one redesigning this, you can now choose between a cheaper non-isolated option or actually implement the isolation, and depending on the connector pinout, connector wiring, and how the target system uses the ground pins, it may or may not be possible to implement the isolation. If it did work fine without isolaton up to this date and no one complained about missing isolation, you might not need it.

You can also fix any other mistakes too if you find out the termination resistors or failsafe bias resistors are not implemented in the system at all.

\$\endgroup\$
3
\$\begingroup\$

Yes it makes sense to isolate one side of the RS485 because this then ensures that earth or ground fault/anomalies that may manifest on the external system are not causing an interfering current to flow through the vulnerable RS485 data wires. In short, it makes sense to me. Of course having isolation at both ends is ideal but, having it at one end is a lot more resilient to having none.

EDIT since a circuit diagram was added to the question

Your circuit diagram uses an isolated RS485 interface but unfortunately does not use it in an isolated way. This is because GND_ISO connects to GND1 (the non isolated side of the converter). So, the only redeeming provision in your circuit is that it is battery powered. However, if your battery powered device also connects to other things (that may be earthed/grounded) then you should be cautious about the design as it currently stands.

\$\endgroup\$
9
  • \$\begingroup\$ Yeah but if there is isolation then it's missing a common reference, so it's a design error. \$\endgroup\$ Commented Jul 29 at 13:37
  • \$\begingroup\$ I'd hardly call the image in the question a design; it's more of a block diagram so, when you downvoted my answer presumably you meant to downvote the question @Justme \$\endgroup\$ Commented Jul 29 at 13:43
  • \$\begingroup\$ I guess more like answering before verifying if the provided info is correct, but you are correct by providing a generic answer for a generic question even if the actual design may still have a horrible error lurking in there. Will remove the downvote. \$\endgroup\$ Commented Jul 29 at 13:49
  • 1
    \$\begingroup\$ Did you remove the dv @Justme || I mean, I didn't see you apply it to the other answer. \$\endgroup\$ Commented Jul 29 at 14:06
  • 1
    \$\begingroup\$ @Justme thanks. \$\endgroup\$ Commented Jul 29 at 19:23
1
\$\begingroup\$

At $6 per chip this is a really expensive way to design in a sacrificial part for external surges and transients.

Perhaps they want to prevent leaking battery when the system is disabled due to stray currents. But the 12V booster design goes against this.

It also isn't the last device on the bus, there are no termination resistors.

A properly enclosed battery powered product may not need isolation.
You can save ~$5.70 bom costs on this device by using a regular transceiver.

Speculation:
This smells like a novice designer using isolation as shotgun approach to solve EMI problems they've created by using a ~1 MHz booster switcher on a poor pcb layout.

\$\endgroup\$
3
  • \$\begingroup\$ The chip isn't even isolating as everything is just wired together. Indeed your speculation may be correct. On the bright side, since it does not isolate, the bus has the required common reference between transceiver chips. \$\endgroup\$ Commented Jul 29 at 19:56
  • \$\begingroup\$ @Justme Yes, I noticed. However it does only reference the potentials. No current can flow. \$\endgroup\$ Commented Jul 30 at 7:01
  • \$\begingroup\$ That is debateable. Current can flow, and since same ground is shared with load return current, it will flow. Difference in the reference potential is allowed, as long as it does not exceed the allowed 7V defined in the RS-485 standard. \$\endgroup\$ Commented Jul 30 at 8:25
1
\$\begingroup\$

With gnd_iso connected to gnd1 you're not getting any isolation. so you're spending money on expensive parts and depleting your batteries at an accelerated rate for no benefit.

what is missing is any surge protection, there's an asymmetical TVS that rhymes with "712" that's probably a good start

\$\endgroup\$
-1
\$\begingroup\$

ADM2587 has an internal dc/dc converter, the power supply for the RS485 transceiver is galvanically isolated from MCU side. So, it is isolated.

\$\endgroup\$
2
  • \$\begingroup\$ It is isolated, no doubt. The point being that simply connecting two data wires to an isolated RS-485 interface is incorrect. \$\endgroup\$ Commented Jul 29 at 13:57
  • \$\begingroup\$ And as it turned out, the chip can isolate but in the schematics it does not. \$\endgroup\$ Commented Jul 29 at 19:52

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.